|
|
IEEE zadržava pravo neobjavljivanja radova u bazi IEEE Xplore ukoliko radovi nisu prezentirani na skupu (odnosi se samo na radove na engleskom jeziku).
| Authors are kindly asked to prepare presentations lasting no more than 10 minutes. |
| Pozvano predavanje |
M. Čubrilo (Faculty of Organization and Informatics, Varaždin, Croatia)
Some Logical and Related Formalisms, Programming Paradigms, and Development Environments for the (New) AI |
| Radovi |
A. Davydov, A. Larionov, N. Nagul (Matrosov Institute for System Dynamics and Control Theory at Siberian Branch of Russian Academy of S, Irkutsk, Russian Federation)
The Formal Logic Approach for Checking the Observability of a Specification Language on DES Functioning 
Using the new approach to the formalization of controlled discrete-event systems (DES), based on positively constructed formulas (PCFs) calculus, the algorithm for testing the observability of the specification languages is presented in this paper. An information mapping in the form of a natural projection and an automata-based representation of a logical DES is considered. Sequences of events, causing changes in the state of the system, are generated as words of a formal language. Discrete-event models of autonomous underwater vehicle (AUV) and AUV groups are developed, which describe the main high-level functions of AUV in surveillance missions. Their formalization in the form of PCFs is presented.
|
R. Čorić, M. Đumić, S. Jelić (J. J. Strossmayer University of Osijek Department of Mathematics, Osijek, Croatia)
A Genetic Algorithm for Group Steiner Tree Problem
In Group Steiner Tree Problem (GST) we are given a weighted undirected graph and family of subsets of vertices which are called groups. Our objective is to find a minimum-weight subgraph which contains at least one vertex from each group (groups do not have to be disjoint). GST is NP-hard combinatorial optimization problem that arises from many complex real-life problems such as finding substrate-reaction pathways in protein networks, progressive keyword search in relational databases, team formation in social networks, etc. Heuristic methods are extremely important for finding the good-enough solutions in short time. In this paper we present genetic algorithm for solving GST. We also give results of computational experiments with comparisons to optimal solutions.
|
E. Zunic (Info Studio d.o.o., Sarajevo, Bosnia and Herzegovina), H. Hasic, K. Hodzic (Faculty of Electrical Engineering, Sarajevo, Bosnia and Herzegovina), S. Delalic, A. Besirevic (Faculty of Sciences and Mathematics, Sarajevo, Bosnia and Herzegovina)
Predictive Analysis based Approach for Optimal Warehouse Product Positioning
Building a successful warehouse management system encompasses solving many problems of different nature to reshape the general workflow and ensure improvements in terms of resource management. In order for such a system to be accepted and used by a logistics company, those solutions need to be presented through a simple, adaptable and most importantly, a feasible software solution. One of the aspects that needs to be covered while building a warehouse management system is the optimal product placement in the warehouse. If the products are strategically placed, all the other improvement strategies like stock to pick zone item transfer and item picking order become more efficient and easier to implement. In this paper, tactics and issues regarding the optimal product placement in a warehouse are analyzed in detail through a real-world case study. Solutions of this problem largely differ for new, empty warehouses and for already operative warehouses not able to carry out a complete stocktaking process. Approaches for both possible situations are proposed and tested out on two different warehouses in a medium-to-large logistics company.
|
S. Lovrenčić, M. Šestak, D. Andročec, D. Plantak Vukovac, Z. Stapić (Svučilište u Zagrebu, Fakultet organizacije i informatike, Varaždin, Croatia)
Upravljanje znanjem i baze znanja u kontakt centrima
Upravljanje znanjem koje je danas potpomognuto modernom tehnologijom važan je čimbenik uspješnosti poslovnih sustava, pa tako i kontakt centara. Poseban izazov u izgradnji baze znanja kontakt centra predstavlja organiziranje i strukturiranje podataka i znanja prikupljenog kroz različite kanale, a potrebno ga je sistematizirati, strukturirati i pohraniti te potom ponovno učiniti dostupnim zaposlenicima (agentima) i korisnicima kontakt centara (također kroz različite kanale). Baza znanja kontakt centra koja uključuje različite tipove podataka, pa i multimedijske podatke, te njezino otvaranje korisnicima ima ulogu osigurati što jednostavniji pristup znanju, informacijama i podacima, i omogućiti korisniku samostalno rješenje problema prije samog poziva agentu kontakt centra. Rad istražuje moguće pristupe upravljanju znanjem u kontakt centrima kroz izgradnju baza znanja. Analiza postojećih istraživanja o korištenju koncepata upravljanja znanjem u kontakt centrima i sustavima za upravljanje odnosima s klijentima (CRM) pokazalo je da je postupak integracije ova dva pristupa još uvijek u početnoj fazi, što otvara znatan prostor za napredak, te su navedeni mogući smjerovi razvoja.
|
A. Yurin, A. Berman, O. Nikolaychuk, N. Dorodnykh (Matrosov Institute for System Dynamics and Control Theory, Siberian Branch of the Russian Academy of, Irkutsk, Russian Federation), M. Grishenko (CentraSib LLC., Irkutsk, Russian Federation)
The Domain-Specific Editor for Rule-Based Knowledge Bases
The aim of the paper is to describe a domain-specific editor for the design of rule-based knowledge bases in the field of the prognosis of technical conditions and remaining operation time of petrochemical equipments. The architecture, main functions and a structure of files for configuration of the editor are presented. The feature of the editor is a semantic layer in the form of a platform-independent model. This layer provides to configure the editor with the account of features of a subjects domain. The semantic layer is implemented as a set of domain specific templates describing facts and rules (cause-and-effect relationships). These templates help to abstract from the syntax of certain knowledge representation languages (programming languages for knowledge bases, in particular, CLIPS - C Language Integrated Production System) and generate the graphic user interface elements.
|
A. Yurin, A. Berman, N. Dorodnykh, O. Nikolaychuk ( Matrosov Institute for System Dynamics and Control Theory of Siberian Branch of Russian Academy of, Irkutsk, Russian Federation), N. Pavlov (CentraSib LLC., Irkutsk, Russian Federation)
Fishbone Diagrams for the Development of Knowledge Bases
The paper describes an approach for the automated development of rule-based knowledge bases by transforming fishbone diagrams. The approach is based on the identification and extraction of structural cause-effect elements of fishbone diagrams and their transformation into the elements of a target knowledge representation language, in particular, C Language Integrated Production System (CLIPS). The source metamodel of fishbone diagrams, the target metamodel for a unified representation of rules (a model for representation logical rules), transformation operators and a transformation technique are presented. An illustrative example describes the development of a rule-based knowledge base for diagnosing and forecasting the states of complex technical systems based on the approach proposed.
|
M. Kenzin, I. Bychkov, N. Maksimkin (Matrosov Institute for System Dynamics and Control Theory of Siberian Branch of Russian Academy of , Irkutsk, Russian Federation)
An Approach to Route Underwater Mobile Robots under Continuous Squad Rotation
Multiple cooperative vehicle systems hold great promise for use in large-scale oceanographic operations due to ability of high-resolution surveying in both time and space. Multi-objective missions of long-duration require underwater robots to recharge their batteries periodically by docking to the specialized underwater bases (resurfacing in case of solar batteries). Furthermore, it should be taken into account, that the real world underwater vehicle systems are partially self-contained and could be subjected to any malfunctions and unforeseen events. Thus, it is a problem of considerable practical interest to effectively route the group of vehicles under continuous rotation. We propose a dynamic rendezvous point-selection scheme based on pre-estimated vehicle rotation cycle and an evolutionary path planner to route the group of robots ensuring well-timed accomplishment of all tasks and simultaneous arrival of vehicles at their selected rendezvous destination.
|
A. Davydov (Matrosov Institute for System Dynamics and Control Theory of Siberian Branch of Russian Academy of , Irkutsk, Russian Federation)
Logic Level of Control for Robot Groups Using the Method of Positively Constructed Formulas
The use of the logic calculus of positively-constructed formulas (PCF) for generating plans of actions for groups of underwater robots is considered in this paper. These actions are supposed to be managed by the high level of some control system and plans can be generated in automatic or interactive mode. An example of high-level control implementation for a group of autonomous underwater robots on the basis of formalization of the subject area using the PCFs is presented.
|
A. Feoktistov, R. Kostromin (Matrosov Institute for System Dynamics and Control Theory of SB RAS, Irkutsk, Russian Federation), A. Tchernykh (CICESE Research Center, Ensenada, Mexico)
Agent Behavior Model for Distributed Computing Management in the Environment with Virtualized Resources
The paper address the actual problem related to the management of jobs generated by scalable applications in an environment that integrate Grid system and cloud infrastructure. The complexity of the management arise due to the existence of differences in the models of cloud and Grid computing, as well as the conflicts between the preferences of environment resource owners and quality criteria for the job management that are defined by users of these resources. Often, the use of a multi-agent system to manage distributed computing makes it possible to achieve a significant success in solving the problem. Agents in such a system are intelligent software entities endowed with rights and responsibilities for servicing computational processes. They represent the interests of resource owners and their users. Agents need to interact between themselves to achieve their goals (ensuring the preferences of resource owners and job execution criteria of resource users) and satisfy their mental properties (for example, degree of intentions to execute jobs of different classes). Providing to an agent all the knowledge and functionality it needs is a non-trivial problem for the known multi-agent systems used to manage distributed computing in the current practice. In this regard, we propose a multi-agent system to manage jobs in a heterogeneous distributed computing environment with cloud resources. It provides the capability to design agents. In this system, we implemented a new specialized model for the agent operation and tools for designing it. The model is based on the integrated use of paradigms of conceptual and finite state machine programming. The advantages of using agents with the proposed model of their behavior are demonstrated by an example of the job management for a scalable application for solving optimization problems of warehouse logistics. In comparison with the known management systems, a significant improvement in the efficiency of solving user problems and environment resource usage is observed when our multi-agent system is used.
|
M. Nadrljanski (Maritime Faculty, Split, Croatia), Đ. Vukić, Đ. Nadrljanski (University College of Inspection and Personnel Management, Split, Croatia)
Multi-Agent Systems in E-Learning
New technologies have influenced the development of e-learning systems, especially in terms of improving the learning process and more efficient knowledge acquisition by adapting to the mental model of a user. One of the solutions for complexity of designing such systems is the integration of multi-agent technology into e-learning systems. A multi-agent system is a group of independent agents which come up with an optimal solution through mutual communication. This paper defines and explains the concepts related to agents and multi-agent systems in e-learning and presents an overview of the research in the context of multi-agent system application in e-learning with emphasis on the types of agents.
|
K. Vidović (Ericsson Nikola Tesla, Zagreb, Croatia), S. Mandžuka, D. Brčić (Faculty of Transport and Traffic Sciences, Zagreb, Croatia)
ANFIS Based Expert System for Urban Mobility Estimation Based on user’s Telecommunication Activities in Public Mobile Network
Urban mobility can be estimated using urban mobility indicators defined and calculated from anonymised data from public mobile communication network. The anonymised data set is derived from Call Data Records database, that contains inputs from user’s telecommunication activities. Data excerpt is used as an input for calculation of urban mobility indicators values, such as origin destination matrix, distance, travel time etc… This article aims at providing a framework for determination of relation between the values of urban mobility indicators and values of mobility estimation. Therefore, a novel approach will be proposed, that utilises an expert system based on the method of fuzzy logic, the ANFIS method which functions on the principle of applying conclusion methods which characterise neural networks with the goal of determining parameters of an indirect conclusion system (Fuzzy Inference System – FIS). The result is ANFIS based expert system framework for urban mobility estimation, which will be trained using set of rules that were determined using method of surveying experts in domain of urban mobility.
|
E. Zunic (Info Studio d.o.o., Sarajevo, Bosnia and Herzegovina), A. Besirevic, S. Delalic (Faculty of Sciences and Mathematics, Sarajevo, Bosnia and Herzegovina), K. Hodzic, H. Hasic (Faculty of Electrical Engineering, Sarajevo, Bosnia and Herzegovina)
A Generic Approach for Order Picking Optimization Process in Different Warehouse Layouts
The warehouse layout directly affects the process of receiving and storing of goods, as well as the order picking process. A standard layout has its palette places arranged in parallel shelves with aligned cross-aisles. However, a big number of warehouses use a shelf layout in which the complete warehouse can’t be represented as a sequence of parallel shelves. Effective calculation of distances between positions in the warehouse presents a significant step towards effective order collection and distribution of goods. This paper describes a generic approach for calculating distances within a warehouse which doesn’t necessarily have standard layout. It describes the application of the algorithm on different warehouse layouts in which the shelves can be split into smaller units so that inside of the each unit the shelves are distributed in a standard way. Dynamic programming was used for the calculation of distances inside of those units. An analysis and testing of the algorithm was performed on two warehouses of middle size with non-standard shelf layout. The algorithm was tested in the process of collecting orders, as well as the process of moving goods from the stock to the pick zone.
|
| Radovi |
M. Horvat, M. Dobrinić, M. Novosel (Zagreb University of Applied Sciences, Zagreb, Croatia), P. Jerčić (Blekinge Institute of Technology, Karlskrona, Sweden)
Assessing Emotional Responses Induced in Virtual Reality Using a Consumer EEG Headset: A Preliminary Report
We report on a pilot study involving emotion elicitation in virtual reality (VR) and assessment of emotional responses with a consumer-grade EEG device. The stimulation used HTC Vive VR system showing pictures from NAPS database within a specifically designed virtual environment. The stimulation consisted of two distinct sequences with 10 pictures of happiness and 10 pictures of fear. Each picture was contained in a separate virtual room that the participants traveled through along a preset path. The estimation employed EMOTIV EPOC+ 14-channel EEG headset and a custom-developed application. The software wirelessly received EEG signals from alpha, beta low, beta high, gamma and theta bands, time-stamped them and dynamically stored in a relational database for subsequent analysis. Our preliminary results show that statistically significant correlations between valence and arousal ratings of pictures and EEG bands are present but highly personalized. Simultaneous correct placement of VR and EEG headsets is demanding and precise localization of electrodes is difficult. In fact, if emotion estimation is not strictly necessary we recommend using devices with fewer electrodes. Nevertheless, we found the EEG to be effective. By acknowledging its limitations, and using the headset in the correct context, experiments involving emotions may be significantly amended.
|
D. Bužić (VSITE, Zagreb, Croatia), J. Dobša (FOI, Varazdin, Croatia)
Lyrics Classification Using Naive Bayes
Text classification is an important and common task in supervised machine learning. The Naive Bayes Classifier is a popular algorithm that can be used for this purpose. The goal of this research was whether the Naive Bayes classifier can successfully predict song performer based solely on lyrics. A dataset that has been created consists of lyrics performed by Nirvana and Metallica, 207 songs in total. Model evaluation measures showed very good results: precision of 0.92, recall of 0.89 and F-measure of 0.91, therefore lyrics classification using Naive Bayes can be considered as successful.
|
M. Ašenbrener Katić, S. Čandrlić, M. Pavlić (Department of Informatics, University of Rijeka, Rijeka, Croatia)
Modeling of Verbs Using the Node of Knowledge Conceptual Framework
The paper analyzes the semantics of verbs in sentences in Croatian and English language. The paper presents a representation of different verb tenses in both languages and shows how the sentences containing these tenses are modeled using the conceptual framework Node of Knowledge (NOK). The NOK conceptual framework is used for formal knowledge representation expressed in text, i.e. to represent the knowledge network.
The paper graphically presents the form of English language verb tenses and their hierarchy.
The paper represents models of using verbs in different verb tenses in sentences of natural language in Croatian and English language. The models are presented using the Formalized Node of Knowledge (FNOK) formalisms. The DNOK (Diagram Node of Knowledge) formalism is also used for the representation. These models are a part of the meta-model of language that is necessary for the development of an intelligent information system.
|
I. Strizrep, A. Sović Kržić, D. Seršić (University of Zagreb Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
Automated Classification of Croatian Traditional Music
Croatian traditional music is rich with different music styles. Four of them are on the UNESCO Representative list of the intangible cultural heritage of humanity: two-part singing and playing in the Istrian scale, Becarac singing and playing from Slavonia, Klapa multipart singing of Dalmatia and Ojkanje singing. Every region of Croatia is represented by different instruments, singing styles, rhythm and dynamics.
This paper describes an automated classification of Croatian traditional music into regions. The regions are defined by historical and geographical factors and music style similarities: Slavonia, central Croatia, Medimurje, Istria and Dalmatia. Each region is presented with 20 typical music songs. A sample of each song lasts for 30 seconds. The primary used features are mel-frequency cepstral coefficients, as well as zero crossing rate and sound volume. Extracted features are used in machine learning. As a result, more than 80% of the songs are correctly classified. The result shows how specific Croatian traditional music is and how important is to preserve it for future generations.
|
M. Burić, M. Pobar, M. Ivašić-Kos (Odjel za informatiku Sveučilište u Rijeci, RIJEKA, Croatia)
Object Detection in Sports Videos
Object detection is commonly used in many computer vision applications. In our case, we need to apply the object detector as a prerequisite for action recognition in handball scenes. Object detection, to be successful for this task, should be as accurate as possible and should be able to deal with a different number of objects of various sizes, partially occluded, with bad illumination and deal with cluttered scenes. The aim of this paper is to provide an overview of the current state-of-the-art detection methods that rely on convolutional neural networks (CNN) and test their performance on custom video sports materials acquired during handball training and matches. The comparison of the detector performance in different conditions will be given and discussed.
|
T. Špoljarić (University of Applied Sciences, Zagreb, Croatia), I. Pavić (Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
Performance Analysis of an Ant Lion Optimizer in Tuning Generators' Excitation Controls in Multi Machine Power System
In this paper an ant lion optimizer algorithm is proposed for tuning excitation controls in multi machine power system. Tuning of excitation controls includes change of parameters in automatic voltage regulator (AVR) and power system stabilizer (PSS). A proposed algorithm is used as a swarm intelligence optimization method for finding a best solution for small signal stability problems. A two area – four machine model (TAFM) with two interconnected transmission lines is considered in this paper. Small signal stability disturbances observed in this paper include automatic re-closure of an interconnection line between two areas, short circuit on an interconnection line, short term load outage in first area, and a load change in first area. Performance of a proposed algorithm is compared with other swarm intelligence algorithms such as particle swarm optimization (PSO), velocity relaxed particle swarm optimization and a salp swarm algorithm (SSA). Objective functions used in performance analysis and comparison include integral of time-weighted absolute error (ITAE) and mean value of time domain transitional process quality indicators.
|
J. Nalić (Computer Science and Information Technology, J.J. Strossmayer University of Osijek, Osijek, Croatia), A. Švraka (Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina)
Importance of Data Pre-processing in Credit Scoring Models Based on Data Mining Approaches
Data Mining has become essential tool for discovery of hidden patterns and information in databases. However, for a Data Mining model to be meaningful and effective, data pre-processing is one of the key factors in successful model preparation. In this paper, we have investigated how data pre-processing affects real dataset when applying Data Mining technique for the purpose of predicting default clients in micro-financing institution. Therefore, several data pre-processing techniques have been described and applied to the dataset. Results are shown and compared for both of the cases with Generalized Linear Model and Decision Tree being the two Data Mining classification algorithms used for Credit Scoring model. It is concluded that Credit Scoring Model is much more accurate and efficient when it is executed on data that has been carefully prepared and pre-processed.
|
B. Dragusha (Anti-Corruption Agency, Prishtina, Kosovo), K. Sylejmani, L. Ahmedi, B. Rexha (University of Prishtina - Faculty of Electrical and Computer Engineering, Prishtina, Kosovo)
Mining Data of Anti-Corruption Institutions to Identify Unusual Growing Trends of Assets of Individuals
Identifying unusual grow of assets of senior public officials can be done by analyzing their data (e.g. capital investments, cash, salaries, etc.), which are usually declared at corresponding national institutions for anti-corruption. Commonly, such agencies collect data for thousands of officials every year. Hence, manual analysis of such a big corpus of data, for a short computation time (i.e. in the range of seconds), is infeasible. Therefore, in this paper, we suggest a solution to this problem, which bases on two algorithms that arise from the field of artificial intelligence. First is the k-means algorithm to group the officials into clusters, which resulted into two clusters: officials that have a normal asset grow (tagged as Best Clusters), and officials with unusual asset grow (aka. Bed Clusters). Second, by using the data of the existing officials, we train a Decision Tree method to divide the officials into two classes, namely the so-called Bad and Best classes respectively. Finally, we use this model to match the new public officials against the existing ones, in order to place them into one of the existing groups/classes. The experiments performed over a data set of 2300 asset declarations show that the proposed approach achieves an accuracy of 86% in comparison to human expert analysis.
|
A. Naumoski, G. Mirceva, K. Trivodaliev, K. Mitreski (Faculty of Computer Science and Engineering, Skopje, Macedonia)
Learning Diatom Ecological Models with Fuzzy Order Data Mining Algorithm
The data mining algorithms allow data scientist to extract useful knowledge from raw measured data. The fuzzy data mining algorithms have several advantages over crisp methods, and they have been used more often to obtain ecological knowledge from ecological data. In this paper, we aim to learn suitable habitat models of the ecological conditions where diatoms can exist in lake ecosystems by using the fuzzy pattern tree algorithm. This algorithm uses several different fuzzy concepts, namely, fuzzy membership functions, similarity metrics and order weighted geometric operator to build predictive ecological model that is able to reveal patterns in ecological data and thus find the suitable diatom ecological conditions. Additionally, we have made experimental evaluation of two similarity metrics that influence the accuracy for both descriptive and predictive models. Later, the results of the models are verified with the known ecological preferences found in the literature. Based on the obtained results, in future we plan to improve the order weighted and similarity metrics and test other membership functions on new ecological data.
|
O. Nikolaychuk, A. Pavlov, A. Stolbov ( Matrosov Institute for System Dynamics and Control Theory of Siberian Branch of Russian Academy of, Irkutsk, Russian Federation)
The Software Platform Architecture for the Component-Oriented Development of Knowledge-Based Systems
The problem of creating a software platform for the automated iterative development of applied knowledge-based systems is considered in the article. Well-known component-based software engineering methodology is utilized as a general approach. The proposed architecture of the software platform includes the management sub-system and a set of problem-oriented components that, in addition to implementing the main functionality, should support the unified platform component interface. The management sub-system provides the ability to interactively define functions of the application based on the combination of the platform components methods via visual programming technique. The list of frequently used functions and features of knowledge-based systems and corresponding platform components is suggested in the article. The data control component implements methods for interacting with a data source. The next component provides the ability to create subject domain model in the ontology form. The rule-based reasoning component provides the ability to create a knowledge base on the top of the obtained domain ontology. The data representation component supplies an automatic creation of elements and forms of the user interface. As an illustrative example, the development process of the knowledge-based system for decision support in the infrastructure logistics domain is presented.
|
Z. Sičanica, Z. Oklopčić (Končar – Power Plant and Electric Traction Engineering Inc., Zagreb, Croatia)
Countywide Natural Gas Consumption Forecast, a Machine Learning Approach
This paper proposes several machine learning models for gas consumption forecast. The consumption data used is for a county in Croatia, and the considered models are decision trees, linear regressors, support vector regression, and neural networks. Most of the models show promising results when compared to similar research. The criterion used to identify the forecast quality is the root of the mean squared error, error being the forecast difference from the expected value. The findings presented in this paper can be used to better understand the way the natural gas is consumed in the county and to create a more sophisticated regressor in the future.
|
S. Kalajdziski, G. Radevski, I. Ivanoska, K. Trivodaliev, B. Risteska Stojkoska (Ss. Cyril and Methodius University, Skopje, Macedonia)
Cuisine Classification Using Recipe’s Ingredients
The purpose of this paper is to explore the linkage between recipe’s ingredients and identification of a cuisine. This has been tackled as a problem of cuisine classification. We will examine various approaches (different machine learning algorithms) for recipes classification based on the recipe’s ingredients. The output will be the recommendation of the classification methodology, i.e. what kind of preprocessing can be done to improve the classification and the performance of several classifiers on the dataset we will be using.
|
| Radovi |
D. Brodic (Technical Faculty in Bor, University of Belgrade, Bor, Serbia), A. Amelio (University of Calabria, Rende (CS), Italy), Z. Milivojevic (College of Applied Technical Sciences, Nis, Serbia)
Classification and Differentiation Between Ijekavian and Ikavian Pronunciation
The paper describes a method for the differentiation and classification between two different types of pronunciation: ijekavian and ikavian. It is based on the coding of written text, which is further represented as an image. This image is subjected to pattern analysis in order to extract the feature vector. Then, the feature vector is differentiated by classification tools. The experiment includes the database of texts in ijekavian and ikavian pronunciation which is subjected to the proposed technique. At the end, the proposed method is compared to the other methods. The obtained results are promising.
|
P. Sillberg, P. Rantanen, J. Soini (Tampere University of Technology, Pori, Finland)
Application for The Analysis and Browsing of Images – Use Case: A Public Photo Archive
Automatic tagging and content-based image retrieval has been studied extensively in the past decades. Regardless, for a long time there were only a very few practical implementations targeted for end-users. Today, there are several commercial products for automatic image annotation, but still, in many cases, the applications have remained closed source, and no comprehensive open source solution exists. This paper presents a real-life use case of an analysis of a publicly available photo archive. The use case is used to illustrate the data format utilized in indexing and searching of the analysis results. Furthermore, the user interface of the designed web application used for browsing the photo content is shown in this paper.
|
E. Mešković, D. Osmanović (Univerzitet u Tuzli, Tuzla, Bosnia and Herzegovina)
A System for Monitoring and Visualization of Big Mobility Data
Efficient real-time monitoring and tracking of mobile objects has recently been a focus of a relatively intense research. Mobile objects produce a huge volume of mobility data arriving in the form of continuous data streams that need to be processed, analyzed and visualized in different information flow processing (IFP) applications. In most cases, this visualization is concerned about redrawing mobile objects locations that represent the result of a continuous query on the map. In this paper, we present a system architecture that enables you to write, construct and pose continuous query in our MobyDick framework prototype for processing big mobility data as spatio-temporal data streams in a cluster computing and to visualize their results on the map. System components are based on the Scala programming language and open source software tools for connection and communication with Apache Flink stream processor which is core of our framework prototype. The system is intended for experienced users familiar with Apache Flink paradigm as well for the beginners who can construct queries from their portions and visualize current locations and trajectories of mobile objects on the map.
|
A. Akagic, E. Buza, S. Omanovic, A. Karabegovic (University of Sarajevo, Sarajevo, Bosnia and Herzegovina)
Pavement Crack Detection Using Otsu Thresholding for Image Segmentation
Pavement cracks are the first signs of structural damage in asphalt pavement surfaces. The oldest method for detection and estimation of pavement cracks is human visual inspection, also known as manual visual inspection. However, using human inspectors is time consuming, very expensive and can pose risks to human safety. Another important negative side is the fact that the task generally requires road to be closed. Hence, automatic prevention and reparation of cracks on asphalt surface pavements is an important task, especially because advanced stages can lead to formation of potholes and can be very costly to repair. In this paper, we proposed new unsupervised method for detection of cracks with gray color based histogram and Ostu's threshold in the pavement image. The method divides input image into four independent equally sized sub-images. For every sub-image, the search for cracks is based on the ratio between Ostu's threshold-value and maximum histogram value. Finally, all sub-images are connected into an output image. The method was tested on dataset which contains different pavement images with very versatile types of cracks. The results showed that the proposed method is very fast and can achieve satisfactory performance especially in the cases of low signal-to-noise ratio.
|
M. Krišto, M. Ivašić-Kos (Odjel za informatiku Sveučilište u Rijeci, RIJEKA, Croatia)
An Overview of Thermal Face Recognition Methods
The popularity of surveillance and access control systems grows as well as a need for better security systems particularly in bad lighting conditions or at night. The aim of a security system is to collect as many details as possible to enable a better recognition of persons. In this paper, a comparison of representative thermal face recognition methods will be given, emphasizing their strengths and weaknesses. Then, trends in the development of surveillance and security systems will be outlined such as fusion of visible and thermal images and use of CNN networks. Also, existing challenges of thermal facial recognition and its applications in a real world will be pointed out.
|
E. Turajlic (Faculty of Electrical Engineering, University of Sarajevo, Sarajevo, Bosnia and Herzegovina)
Application of Firefly and Bat Algorithms to Multilevel Thresholding of x-Ray Images
Multilevel image thresholding is a challenging digital image processing problem with numerous applications, including image segmentation, image analysis and higher level image processing. Although, threshold estimation based on exhaustive search is a relatively straight forward task, it can be computationally very expensive to evaluate optimal thresholds when the number of threshold levels is large. In this paper, a metaheuristic approach to multilevel thresholding of x-ray images has been examined. Specifically, firefly and bat algorithms are used in the conjunction with Kapur’s entropy, Tsallis entropy and Otsu’s between-class variance criterion to estimate optimal threshold values. The performance of various image segmentation strategies have been evaluated on a dataset of x-ray images. The simulation results show that the bat algorithm in conjunction with Otsu’s objective function offers the best X-ray image segmentation strategy. Out of all considered strategies, this multilevel thresholding approach to image segmentation produces the highest PSNR and SSIM values as well as fast execution times.
|
A. Radovan, Ž. Ban (Fakultet elektrotehnike i računarstva, Zagreb, Croatia)
Prediction of HSV Color Model Parameter Values of Cloud Movement Picture Based on Artificial Neural Networks
In order to predict the exact moment of Sun shading by clouds and Sun cover duration to optimize the energy flow in the microgrid with a solar photoelectric system, it is essential to transform cloud images from RGB color model into HSV color model to be able to precisely detect cloud edges and centroids for prediction of their movements. Parameters that define the quality of the processed image depend on the range of values for Hue, Saturation, and Value (HSV) components and the size of structural element used by morphological operations erosion and dilatation. The dynamics of clouds and changing their shapes, sizes, and colors require constant adjustments of those parameters by a human to get best results. This paper deals with prediction and automatic setting of the HSV parameters by using an artificial neural network and supervised learning. The image processing and parameters prediction were performed by an application developed in Java programming language based on JavaCV library and Encog framework for implementation of the artificial neural network.
|
I. Tomičić, P. Grd, M. Bača (Fakultet organizacije i informatike, Varaždin, Croatia)
A Review of Soft Biometrics for IoT
The Internet of Things (IoT) can be defined as everyday physical objects being connected to the internet and being able to identify themselves to other devices. In recent years the Internet of Things (IoT) was identified as one of the emerging technologies. Research in this area has increased considerably and future research will have to include other technologies such as biometrics to complement the development of IoT systems. The idea of this paper is to give an overview of biometric characteristics applicable to IoT with emphasis on soft biometric characteristics and possible application scenarios in IoT.
|
Z. Kljaić (Ericsson Nikola Tesla d.d., Zagreb, Croatia), E. Briški (HAKOM, Croatian Regulatory Authority for Network Industries, Zagreb, Croatia), H. Vojvodić (Auto Hrvatska d.d., Zagreb, Croatia), N. Amin (Ericsson Nikola Tesla d.d., Zagreb, Croatia)
Benefits of Utilisation of GPS Error Mitigation Models for Intelligent Transport Systems
This manuscript outlines the research results of potential pitfalls in utilisation due to inadequate or inappropriate usage of the GPS satellite navigation system for Intelligent Transport Systems. The impact of the error correction models and techniques on the GPS positioning performance is addressed through analysis of three use-case studies conducted on a GPS software-defined radio fed with experimetally collected GPS pseudoranges. Recommendations are proposed in regard to utilisation of the standard error mitigation procedures for Intelligent Transport Systems.
|
M. Milenkoski, K. Trivodaliev, S. Kalajdziski, M. Jovanov, B. Risteska Stojkoska (Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Skopje, Macedonia)
Real Time Human Activity Recognition on Smartphones Using LSTM Networks
In this paper, we developed new method for activity recognition using smartphone accelerometers data. Our method is based on Long Short Term Memory (LSTM) networks, as a special type of neural networks who can remember information from further back in the past. We tested our method on publicly available dataset collected under laboratory conditions and confirm that our method has almost equal performance with other well- known reference algorithms from the literature. The advantage of our method is that it works directly with the raw accelerometer data, and completely bypasses the process of generating hand-crafted features. Additionally, we tested out method on field data and compared the results with laboratory obtained dataset. The results show that, for most of the activities, our method performs only a few percentage points worse than on labdata. This is a very important conclusion as it emphasis that smartphones can be effectively used for accurate activity recognition, which can save resources in many public health studies that investigate physical activity of the population.
|
M. Petkov, B. Risteska Stojkoska, S. Kalajdziski, I. Ivanoska, K. Trivodaliev (Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, Skopje, Macedonia)
Intelligent Analysis of Economic Parameters of European Countries
Engineers today are often bound to work with complex data, with valuable knowledge to be extracted and models to be built for predicting future behavior. With a given set of various economic parameters for the European countries, the goal of this paper is to analyze communities of countries based on the predefined parameters. The first step is the analysis and creation of the corresponding graphs as the most adequate data representation. Graph structure has the benefit of simplicity, but also the power to capture multiple relations and dependencies among the entities of interest. The preprocessing steps for creating a graph include calculating correlation coefficients for the parameters, which are used for weighing the graph. The weighted graphs are then analyzed for their underlying structure using clustering algorithms which produce various communities depending on the settings employed. In this paper, the focus was on two clustering algorithms, namely Louvain and Edge Betweenness, as representatives of two essentially different approaches in community detection. Experiments are performed using different correlation calculations for graph weighing and different settings for cluster extraction and the results are analyzed in terms of the quality of clusters produced, both in terms of modularity score and economic meaning.
|
Z. Balaž (Tehničko veleučilište Zagreb, Zagreb, Croatia), B. Balaž (Rudarsko-geološko-naftni fakultet, Zagreb, Croatia)
Inteligentni sustavi budućnosti za aplikativnu tribologiju
Tribologija kao interdisciplinarna, obuhvaća prvenstveno strojarstvo, metalurgiju, kemiju i fiziku. U današnjem, izotropnom dobu, područje njezine glavne primjene su mehaničke konstrukcije i materijali čije je ponašanje predvidivo. U skoroj budućnosti očekuje se primjena anizotropnih materijala u području tribologije, koji će postati „pametni materijali“ koji „surađuju“ i pridonose poboljšanju stanja na mjestu gdje su ugrađeni. U radu je prikazano na idejnoj razini buduće rješenje zaštite broda u kojem pametni materijali surađuju i uklanjaju koroziju i druge štetne tvari nataložene na oplatu te time zamjenjuju ljude na održavanju i štede vrijeme i novac. Nanosenzori kao komponente novih mikro–elektro-mehaničkih sustava, (MEMS-a), uz pomoć inteligentnih sustava kognitivne kibernetike, preuzimaju, pohranjuju i prosljeđuju relevantne podatke uz pomoć kojih je moguće trenutno djelovanje, a s obzirom na svojstva anizotropnosti materijala koji se primjenjuju i koncepta „big data“ moguće je učenje za budući razvoj.
|
|
Osnovni podaci:
Voditelji:
Slobodan Ribarić (Croatia), Andrea Budin (Croatia)
Predsjednik Međunarodnog programskog odbora:
Karolj Skala (Croatia)
Međunarodni programski odbor:
Enis Afgan (Croatia), Slaviša Aleksić (Germany), Slavko Amon (Slovenia), Lene Andersen (Denmark), Vesna Anđelić (Croatia), Michael E. Auer (Austria), Dubravko Babić (Croatia), Snježana Babić (Croatia), Almir Badnjevic (Bosnia and Herzegovina), Marko Banek (Croatia), Mirta Baranović (Croatia), Bartosz Bebel (Poland), Ladjel Bellatreche (France), Petar Biljanović (Croatia), Eugen Brenner (Austria), Ljiljana Brkić (Croatia), Gianpiero Brunetti (Italy), Marian Bubak (Poland), Andrea Budin (Croatia), Željko Butković (Croatia), Željka Car (Croatia), Jesús Carretero Pérez (Spain), Matjaž Colnarič (Slovenia), Alfredo Cuzzocrea (Italy), Marina Čičin-Šain (Croatia), Marko Čupić (Croatia), Davor Davidović (Croatia), Marko Delimar (Croatia), Saša Dešić (Croatia), Todd Eavis (Canada), Maurizio Ferrari (Italy), Tiziana Ferrari (Netherlands), Bekim Fetaji (Macedonia), Nikola Filip Fijan (Croatia), Renato Filjar (Croatia), Tihana Galinac Grbac (Croatia), Enrico Gallinucci (Italy), Dragan Gamberger (Croatia), Paolo Garza (Italy), Liljana Gavrilovska (Macedonia), Ivan Gerlič (Slovenia), Matteo Golfarelli (Italy), Stjepan Golubić (Croatia), Montserrat Gonzales (United Kingdom), Francesco Gregoretti (Italy), Stjepan Groš (Croatia), Niko Guid (Slovenia), Jaak Henno (Estonia), Ladislav Hluchy (Slovakia), Željko Hocenski (Croatia), Vlasta Hudek (Croatia), Darko Huljenic (Croatia), Željko Hutinski (Croatia), Robert Inkret (Croatia), Mile Ivanda (Croatia), Hannu Jaakkola (Finland), Matej Janjić (Croatia), Leonardo Jelenković (Croatia), Rene Jerončić (Croatia), Dragan Jevtić (Croatia), Admela Jukan (Germany), Robert Jones (Switzerland), Peter Kacsuk (Hungary), Aneta Karaivanova (Bulgaria), Tonimir Kišasondi (Croatia), Marko Koričić (Croatia), Tomislav Kosanović (Croatia), Dieter Kranzlmüller (Germany), Marko Lacković (Croatia), Erich Leitgeb (Austria), Maria Lindén (Sweden), Dražen Lučić (Croatia), Marija Marinović (Croatia), Ludek Matyska (Czech Republic), Mladen Mauher (Croatia), Igor Mekjavic (Slovenia), Igor Mekterović (Croatia), Branko Mikac (Croatia), Veljko Milutinović (Serbia), Nikola Mišković (Croatia), Vladimir Mrvoš (Croatia), Jadranko F. Novak (Croatia), Predrag Pale (Croatia), Jesus Pardillo (Spain), Nikola Pavešić (Slovenia), Branimir Pejčinović (United States), Dana Petcu (Romania), Juraj Petrović (Croatia), Damir Pintar (Croatia), Željka Požgaj (Croatia), Slobodan Ribarić (Croatia), Janez Rozman (Slovenia), Rok Rupnik (Slovenia), Dubravko Sabolić (Croatia), Zoran Skočir (Croatia), Ivanka Sluganović (Croatia), Mario Spremić (Croatia), Vlado Sruk (Croatia), Stefano Stafisso (Italy), Uroš Stanič (Slovenia), Ninoslav Stojadinović (Serbia), Jadranka Šunde (Australia), Aleksandar Szabo (Croatia), Laszlo Szirmay-Kalos (Hungary), Davor Šarić (Croatia), Dina Šimunić (Croatia), Zoran Šimunić (Croatia), Dejan Škvorc (Croatia), Velimir Švedek (Croatia), Antonio Teixeira (Portugal), Edvard Tijan (Croatia), A Min Tjoa (Austria), Roman Trobec (Slovenia), Sergio Uran (Croatia), Tibor Vámos (Hungary), Mladen Varga (Croatia), Marijana Vidas-Bubanja (Serbia), Mihaela Vranić (Croatia), Boris Vrdoljak (Croatia), Slavomir Vukmirović (Croatia), Yingwei Wang (Canada), Mario Weber (Croatia), Roman Wyrzykowski (Poland), Damjan Zazula (Slovenia)
Prijava/Kotizacija:
|
PRIJAVA / KOTIZACIJE
|
CIJENA U EUR-ima
|
|
Prije 7.5.2018.
|
Poslije 7.5.2018.
|
| Članovi MIPRO i IEEE |
180
|
200
|
| Studenti (preddiplomski i diplomski studij) te nastavnici osnovnih i srednjih škola |
100
|
110
|
| Ostali |
200
|
220
|
Popust se ne odnosi na studente doktorskog studija.
Kontakt:
Slobodan Ribarić
Fakultet elektrotehnike i računarstva
Unska 3
10000 Zagreb, Hrvatska
Tel.:+385 1 612 99 52
Fax: +385 1 612 96 53
E-mail: slobodan.ribaric@fer.hr
Andrea Budin
Ericsson Nikola Tesla d.d.
R&D Center
Krapinska 45
10000 Zagreb, Hrvatska
Tel.:+385 1 365 34 23
Fax: +385 1 365 3548
E-mail: andrea.budin@ericsson.com
Najbolji radovi bit će nagrađeni.
Prihvaćeni radovi bit će objavljeni u zborniku radova s ISBN brojem. Radovi napisani na engleskom jeziku i prezentirani na skupu bit će poslani za objavljivanje u bazi IEEE Xplore.
Postoji mogućnost da se odabrani znanstveni radovi uz određenu doradu objave u međunarodnom časopisu Journal of Computing and Information Technology (CIT).
Predsjednik Međunarodnog programskog odbora:
Karolj Skala (Croatia)
Međunarodni programski odbor:
Enis Afgan (Croatia), Slaviša Aleksić (Germany), Slavko Amon (Slovenia), Lene Andersen (Denmark), Vesna Anđelić (Croatia), Michael E. Auer (Austria), Dubravko Babić (Croatia), Snježana Babić (Croatia), Almir Badnjevic (Bosnia and Herzegovina), Marko Banek (Croatia), Mirta Baranović (Croatia), Bartosz Bebel (Poland), Ladjel Bellatreche (France), Petar Biljanović (Croatia), Eugen Brenner (Austria), Ljiljana Brkić (Croatia), Gianpiero Brunetti (Italy), Marian Bubak (Poland), Andrea Budin (Croatia), Željko Butković (Croatia), Željka Car (Croatia), Jesús Carretero Pérez (Spain), Matjaž Colnarič (Slovenia), Alfredo Cuzzocrea (Italy), Marina Čičin-Šain (Croatia), Marko Čupić (Croatia), Davor Davidović (Croatia), Marko Delimar (Croatia), Saša Dešić (Croatia), Todd Eavis (Canada), Maurizio Ferrari (Italy), Tiziana Ferrari (Netherlands), Bekim Fetaji (Macedonia), Nikola Filip Fijan (Croatia), Renato Filjar (Croatia), Tihana Galinac Grbac (Croatia), Enrico Gallinucci (Italy), Dragan Gamberger (Croatia), Paolo Garza (Italy), Liljana Gavrilovska (Macedonia), Ivan Gerlič (Slovenia), Matteo Golfarelli (Italy), Stjepan Golubić (Croatia), Montserrat Gonzales (United Kingdom), Francesco Gregoretti (Italy), Stjepan Groš (Croatia), Niko Guid (Slovenia), Jaak Henno (Estonia), Ladislav Hluchy (Slovakia), Željko Hocenski (Croatia), Vlasta Hudek (Croatia), Darko Huljenic (Croatia), Željko Hutinski (Croatia), Robert Inkret (Croatia), Mile Ivanda (Croatia), Hannu Jaakkola (Finland), Matej Janjić (Croatia), Leonardo Jelenković (Croatia), Rene Jerončić (Croatia), Dragan Jevtić (Croatia), Admela Jukan (Germany), Robert Jones (Switzerland), Peter Kacsuk (Hungary), Aneta Karaivanova (Bulgaria), Tonimir Kišasondi (Croatia), Marko Koričić (Croatia), Tomislav Kosanović (Croatia), Dieter Kranzlmüller (Germany), Marko Lacković (Croatia), Erich Leitgeb (Austria), Maria Lindén (Sweden), Dražen Lučić (Croatia), Marija Marinović (Croatia), Ludek Matyska (Czech Republic), Mladen Mauher (Croatia), Igor Mekjavic (Slovenia), Igor Mekterović (Croatia), Branko Mikac (Croatia), Veljko Milutinović (Serbia), Nikola Mišković (Croatia), Vladimir Mrvoš (Croatia), Jadranko F. Novak (Croatia), Predrag Pale (Croatia), Jesus Pardillo (Spain), Nikola Pavešić (Slovenia), Branimir Pejčinović (United States), Dana Petcu (Romania), Juraj Petrović (Croatia), Damir Pintar (Croatia), Željka Požgaj (Croatia), Slobodan Ribarić (Croatia), Janez Rozman (Slovenia), Rok Rupnik (Slovenia), Dubravko Sabolić (Croatia), Zoran Skočir (Croatia), Ivanka Sluganović (Croatia), Mario Spremić (Croatia), Vlado Sruk (Croatia), Stefano Stafisso (Italy), Uroš Stanič (Slovenia), Ninoslav Stojadinović (Serbia), Jadranka Šunde (Australia), Aleksandar Szabo (Croatia), Laszlo Szirmay-Kalos (Hungary), Davor Šarić (Croatia), Dina Šimunić (Croatia), Zoran Šimunić (Croatia), Dejan Škvorc (Croatia), Velimir Švedek (Croatia), Antonio Teixeira (Portugal), Edvard Tijan (Croatia), A Min Tjoa (Austria), Roman Trobec (Slovenia), Sergio Uran (Croatia), Tibor Vámos (Hungary), Mladen Varga (Croatia), Marijana Vidas-Bubanja (Serbia), Mihaela Vranić (Croatia), Boris Vrdoljak (Croatia), Slavomir Vukmirović (Croatia), Yingwei Wang (Canada), Mario Weber (Croatia), Roman Wyrzykowski (Poland), Damjan Zazula (Slovenia)
Mjesto održavanja:
Opatija, sa 170 godina dugom turističkom tradicijom, vodeće je ljetovalište na istočnoj strani Jadrana i jedno od najpoznatijih na Mediteranu. Ovaj grad aristokratske arhitekture i stila već 170 godina privlači svjetski poznate umjetnike, političare, kraljeve, znanstvenike, sportaše, ali i poslovne ljude, bankare, menadžere i sve kojima Opatija nudi svoje brojne sadržaje.
Opatija svojim gostima nudi brojne komforne hotele, odlične restorane, zabavne sadržaje, umjetničke festivale, vrhunske koncerte ozbiljne i zabavne glazbe, uređene plaže i brojne bazene i sve što je potrebno za ugodan boravak gostiju različitih afiniteta.
U novije doba Opatija je jedan od najpoznatijih kongresnih gradova na Mediteranu, posebno prepoznatljiva po međunarodnim ICT skupovima MIPRO koji se u njoj održavaju od 1979. godine i koji redovito okupljaju preko tisuću sudionika iz četrdesetak zemalja. Ovi skupovi Opatiju promoviraju u nezaobilazan tehnološki, poslovni, obrazovni i znanstveni centar jugoistočne Europe i Europske unije općenito.
Detaljnije informacije se mogu potražiti na www.opatija.hr i www.visitopatija.com.
|
|